Sec. ii] FROM CHIAO-WAN-CH`ÊNG TO SHIH-ÊRH-TUN 379
The purpose served by its construction at once become clear when I found that Ch`iao-wanch`êng guarded the approach to the river of a route leading across the Pei-shan ` Gobi ' from Hâmi, on the most direct line between this oasis and Su-chou. It is a route well suited for traffic with camels and horses, and was followed in 1898 by Professor K. Futterer, who has carefully mapped and described it.3 As long as the castrum held its garrison, it must necessarily have been used as the place where caravans when leaving for, or coming from, the Hâmi side took their supplies, refitted, & c. ; for its position on the river and close to cultivated ground on the other side made it obviously a most convenient bridge-head, as it were, for the journey across the desert northwestwards.
It was to serve the needs of the traffic moving along this route as well as to shelter those who ministered to the wants of the garrison, that the suburb of Sarais, booths, &c., had grown up which I found extending, all in ruins, round the western and southern walls of the castrum, as shown by the plan, Pl. 15. The irregular shape and comparative weakness of the enceinte protecting this suburb seemed to indicate that its growth had been gradual and anyhow had not been provided for in the original plan of the station.
- Since the little town was destroyed and abandoned, the reasons which made Ch`iao-wanch`êng the southern terminus for the desert route had for the most part ceased. On this account those who use it now prefer to do the last march from Hâmi on a track which takes them slightly farther to the east. It allows them to strike the river at a point about five miles farther up, where the bare gravel belt on its right bank gives place to a sandy area with abundant vegetation suitable for grazing, as seen in Map No. 40. c. 4. At Ch`iao-wan-ch`êng only the scantiest grazing is obtainable. I subsequently found this new track, which Professor Futterer's guides also had followed, clearly marked at T. XLI. k where it crosses the agger of the Limes.4
There was much in the appearance of the ruined town and in its position as the terminal guard-station of a desert route to bring vividly to mind the Lou-lan station L.A., where the conditions, it is true, are of far more advanced decay. This impression was much strengthened by the examination of the ruined structures scattered over the open ground to the north of the circumvallation. The fact that the line of the ancient Limes actually crossed this ground naturally invested it with additional antiquarian interest. On a massive clay base about 15o yards from the northern gate rose the ruins of a large temple. It was so completely shattered that the interior plan of it could be recovered only by regular excavation. A number of small shrines to the east of it, which attested, perhaps, the piety of individual officers, &c., of the garrison, had suffered far less.
Going another hundred yards to the north I found the Limes agger running here in a perfectly straight continuous line across the bare gravel soil and still rising to a height of 3 or 4 feet in most places. A short distance beyond it I came upon two shallow parallel ditches, with a ruined pavilion by their side. They would have presented a puzzle to me, as they well might to some
3 See Futterer, Geographische Skizze der 11'üste Gobi zwischen Hami and Su-tschôu, Petermanns Mitteilungen, Ergänzungsheft No. x39, 1902. The map by Dr. B. Hassen-stein accompanying the paper and embodying Professor Futterer's very careful compass survey of the route is on the scale of r : 500,000 (not, as the title by a draughtsman's mistake states, of r : I,000,000).
4 It is due to this change of route on the last march that Professor Futterer did not notice Ch`iao-wan-ch`êng, though he must have passed within four miles or so of it. At the time of his journey over this ground (May 25th) visibility
was bound to be very low owing to the prevailing dust-haze of the season. But the mention he makes, loc. cit., p. 23, of a reported ` old monastery ', deserted except for a few monks, evidently refers to it, though his guides seem to have wrongly placed the ruined site on the left bank of the river.
I may note here that if on Professor Futterer's map the perfectly straight direction of the initial portion of his route from Camp xiv were continued right down to the Su-to-ho, without the subsequent bend to the south-south-east, the route would abut on the river at a point exactly corresponding in bearing and distance to Ch`iao-wan-ch`êng.